#MemoryAiProvider

#Import

1import { MemoryAiProvider } from "alepha/ai";

#Overview

The in-memory AI provider: the default everywhere, and the one specs drive.

Each script call queues one response, played back by the next chat call in order. With nothing queued it answers one text reply saying no vendor module is registered, so a dev page with a chat renders and names its own cause instead of failing.

Every request is kept in requests, so a spec can assert what the model would have been sent.